The assistant has a few properties configured to extract from the transcript. Sometimes the assistant does a good job and extracts everything, but sometimes the analysis is empty, and I can see in the call logs that attempts to request LLM models are failing with timeouts. For instance, call log: https://dashboard.vapi.ai/calls/019a0bd9-9d33-7dd7-9ccc-365a76acf805 Looks like we've missed a few important calls because of that. Any ideas what that might be and how we can fix it? https://cdn.discordapp.com/attachments/1430840965127471197/1430840965639307275/example_of_the_log.jpg?ex=68fb3dfd&is=68f9ec7d&hm=0e6bda86b6665a00b5e7ac25d6e1313cfaa05072518ed637e6d5c3d5713efe13&

To address the LLM timeout errors and empty structured data, try increasing the

timeoutSeconds

property in your assistant's model configuration. For example:

json
{
  "model": {
    "provider": "custom-llm",
    "timeoutSeconds": 120,  // Set between 20 and 600 seconds as needed
    "url": "your-endpoint"
  }
}

This should help prevent timeouts during structured data extraction and reduce missed analyses. For more details, see the documentation on the new

timeoutSeconds

property. Source: - [New timeoutSeconds Property in Custom LLM Model](https://docs.vapi.ai/changelog/2025/3/17)

@User We are not using a custom LLM. The assistant is configured to use OpenAI with a custom key.

Hi Relaximus, If you're looking to create or update a structured output, here’s a quick guide: To Create a Structured Output 1. **API Endpoint**: POST

https://api.vapi.ai/structured-output

2. **Headers**: Include

Authorization

with your API Key. 3. **Request Body**: - `name`: Your output's name. - `schema`: Define using JSON Schema specifications ([learn more](https://json-schema.org/)). - Example Schema:

json
     {
       "name": "Support Ticket",
       "schema": { 
         "type": "object",
         "properties": {
           "customer": {
             "type": "object",
             "properties": {
               "name": {"type": "string", "description": "Customer full name"},
               "email": {"type": "string", "format": "email", "description": "Customer email"}
             },
             "required": ["name"]
           }
         },
         "required": ["customer"]
       }
     }

- For more detailed instructions, you can refer to the [API reference](https://docs.vapi.ai/api-reference/structured-outputs/structured-output-controller-create). To Update a Structured Output 1. **API Endpoint**: PATCH

https://api.vapi.ai/structured-output/{id}

2. **Headers**: Include

Authorization

with your API Key. 3. **Path Parameter**:

id

of the structured output you want to update. 4. **Query Parameter**:

schemaOverride

- new schema in JSON format. 5. For detailed steps, you may check the [update documentation](https://docs.vapi.ai/api-reference/structured-outputs/structured-output-controller-update). These endpoints allow you to manage your structured outputs effectively. Feel free to explore additional real-world examples and quickstart guides for structured outputs to gain deeper insights and enhance your implementation.